Abstract
Aiming at both compression and content protection, we propose a context intra-coding scheme for securing surveillance videos. We also develop methods to detect moving objects, skin color and human face in compression domain. In the proposed context intra-coding scheme, every DCT block is identified as either a parent block or a child block. The child block is then represented in terms of the corresponding parent block and the residual. A signature block is further employed to encrypt the residual. Finally, the parent blocks and the encrypted residuals are coded using conventional intra-coding. The frame content is well protected and can be accurately decoded only with the correct secret key. Experiment results demonstrate that our scheme is capable of protecting video content and also detecting significant objects in compression domain without violating personal privacy. © 2011 IEEE.